DESCRIPTION: (Applicant's Description) Regulation of mRNA transcription is one of the major determinants of cellular phenotype. Recent genome-wide expression studies establish that cancerous cells display global alterations in transcript abundance that i) determine neoplastic behavior and ii) predict clinical course and outcome. Here we describe the fIrst intelligent, scaleable, and automated approach to identifying the broader biological significance of these data. Specifically, these methods computationally detect the altered regulation of components of biological pathways in large-scale expression data, using a knowledge base of information about gene function. In Phase I of this grant, we will: Populate a functional genetic knowledge base with more than 45,000 published facts on at least 200 genes involved in two well-established neoplastic subprocesses (programmed cell death and the mitotic cell cycle). Develop two algorithms that will identify functionally related subsets of these genes from standard expression data. Evalute the ability of these algorithms to detect biologically meaningful clusters of genes within I) the complete set of 200 genes in our knowledge base and ii) differentially regulated genes from a limited set of cancer-related expression data. PROPOSED COMMERCIAL APPLICATION: NOT AVAILABLE